Invasive and Metastatic Properties of Advanced Prostate Cancer
Karin Jennbacken 2009
Department of Urology
Lundberg Laboratory for Cancer Research Institute of Clinical Sciences
Sahlgrenska Academy at University of Gothenburg Sweden
Front cover: A normal mouse prostatic gland invaded with tumor cells, in DAPI staining.
ISBN 978-91-628-7761-3
© 2009 Karin Jennbacken
Printed by Intellecta Infolog AB, Västra Frölunda, Sweden
Till Mamma & Pappa
Karin Jennbacken
Department of Urology, Lundberg Laboratory for Cancer Research, Institute of Clinical Sciences, Sahlgrenska Academy at University of Gothenburg, Sweden
ABSTRACT
Prostate cancer is initially androgen-dependent (AD) and therefore androgen deprivation therapy (ADT) is generally used to treat advanced prostate cancer. However, the long-term treatment effects are insufficient and over time an androgen-independent (AI) tumor relapses, which is generally highly aggressive and metastatic. Treatment regimens in the AI stage are only palliative and median patient survival is less than a year. Therefore, new treatment concepts are urgently needed. The purpose of this thesis was to investigate molecular and cellular characteristics of advanced prostate cancer. The specific focus was on characteristics related to invasivity and metastatic ability in the AI stage. An experimental model system comprising of an AD and an AI prostate cancer cell line was used for in vitro studies in cell culture and in vivo studies in immunodeficient mice. In addition, samples from prostate cancer patients were included in the studies and evaluated by immunohistochemical analyses. Studies performed using the experimental model showed that transition into androgen-independency was associated with several prometastatic alterations, including increased migration and tumor cell invasivity into blood vessels. Further, the AI tumors displayed elevated levels of N-cadherin, matrix metalloproteinase 9 (MMP-9) and membrane type-1(MT1)-MMP and decreased expression of the tumor suppressor E-cadherin compared to the AD tumors. Further studies demonstrated that intraprostatic AI tumors were suppressed when grown in intact mice compared to castrated mice, probably by androgen-regulated factors secreted from the prostatic stromal cells. In addition, the proinvasive factor N-cadherin was increased by androgen deprivation in experimental AI tumors and in samples from human prostate cancer. Similarly, N-cadherin was increased in specimens from AI prostate tumors compared to early non-treated tumors and was associated with Gleason score and metastasis. Finally, the results show that the lymphangiogenic factor vascular endothelial growth factor C (VEGF-C) and its receptor VEGFR-3 were elevated in primary tumors from patients with regional lymph node metastases compared to patients without lymph node metastases. In summary, this thesis shows that androgen deprivation and the subsequent development of AI tumors are associated with several prometastatic alterations in the prostate cancer cells. The results also suggest that AI tumors do not thrive in the prostatic environment and supports previous observations of frequent progression of AI prostate cancer as metastases in patients. Moreover, the results indicate a possible role for VEGF-C and N-cadherin in promoting dissemination of tumor cells to distant sites. Thus, N-cadherin and VEGF-C might be potential therapeutic targets for future anti-metastatic treatment for advanced prostate cancer.
Key words: Prostate cancer; Androgen-independent; Castration-resistant; Metastasis, Invasion;
Lymphangiogenesis; Cell adhesion; N-cadherin; VEGF-C; MRI
This thesis is based on the following papers, which will be referred to in the text by their roman numerals:
I. Jennbacken K., Vallbo C., Wang W., Damber JE.
Expression of vascular endothelial growth factor C (VEGF-C) and VEGF receptor-3 in human prostate cancer is associated with regional lymph node metastasis. The Prostate. 2005 Oct 1;65(2):110-116
II. Jennbacken K., Gustavsson H., Welén K., Vallbo C., Damber JE.
Prostate cancer progression into androgen independency is associated with alterations in cell adhesion and invasivity. The Prostate. 2006 Nov 1;66(15):1631-1640
III. Jennbacken K., Gustavsson H., Tešan T., Horn M., Vallbo C., Welén K., Damber JE.
The prostatic environment suppresses growth of androgen-independent prostate cancer xenografts: An effect influenced by testosterone. The Prostate 2009. In press
IV. Jennbacken K., Tešan T., Wang W., Gustavsson H., Damber JE., Welén K.
N-cadherin increases after androgen deprivation and is associated with metastasis in prostate cancer. In manuscript
ABBREVIATIONS... 1
INTRODUCTION ... 2
The normal prostate gland ... 2
Anatomy and physiology... 2
Morphology ... 2
Regulation of the prostate gland... 4
Effects of androgen deprivation ... 5
Prostate cancer ... 5
General background... 5
Prostate carcinogenesis... 6
Diagnosis and pathology ... 7
Treatment of prostate cancer ... 9
Mechanisms of development of AI prostate cancer ... 10
Interactions between cancer cells and stroma ... 13
Invasion and metastasis ... 14
The metastatic process... 14
The seed and soil theory... 16
Tumor dormancy ... 17
Lymphangiogenesis and metastasis... 17
Cell adhesion molecules in cancer ... 20
Role of MMPs for metastasis ... 24
AIMS OF THE THESIS ... 26
MATERIALS AND METHODS... 27
RESULTS AND COMMENTS ... 36
Paper I... 36
Paper II... 37
Paper III ... 38
Paper IV ... 39
GENERAL DISCUSSION ... 41
Characteristics of human AI prostate cancer... 41
Animal models of prostate cancer ... 42
MR as an imaging tool for monitoring tumor growth in mice ... 43
Transition into androgen-independency is associated with prometastatic properties ... 43
AI tumors are suppressed in the prostatic microenvironment... 45
Prometastatic alterations induced by ADT... 46
The role of the lymphatic system for metastasis ... 48
CONCLUSIONS ... 51
POPULÄRVETENSKAPLIG SAMMANFATTNING ... 52
ACKNOWLEDGEMENTS ... 54
REFERENCES... 57
ABBREVIATIONS
ADAMTS1 ADAM metallopeptidase with thrombospondin type 1, motif 1
AD Androgen-dependent
ADT Androgen deprivation therapy
AI Androgen-independent
AR Androgen receptor
ARE Androgen response element BPH Benign prostatic hyperplasia CAF Carcinoma associated fibroblast CAM Cell adhesion molecule
DHT Dihydrotestosterone DLP Dorsolateral prostate ECM Extracellular matrix
EGF Epidermal growth factor EMT Epithelial mesenchymal transition FBS Fetal bovine serum
FBS-DCC Fetal bovine serum - dextran charcoal treated FGF Fibroblast growth factor
FGFR Fibroblast growth factor receptor GnRH Gonadotropin releasing hormone IGF Insulin-like growth factor
KGF Keratinocyte growth factor LH Luteinizing hormone
LNCaP Lymph node carcinoma of the prostate MMP Matrix metalloproteinase
MRI Magnetic resonance imaging
MT1-MMP Membrane type 1-matrix metalloproteinase
MVD Microvessel density
NE Neuroendocrine
NRCAM Neuronal cell adhesion molecule PCDH20 Protocadherin 20
PDGF Placental derived growth factor PIN Prostatic intraepithelial neoplasia PSA Prostate specific antigen
RGS2 Regulator of G-protein signaling 2, 24 kDa RT-PCR Reverse transcriptase polymerase chain reaction SCID Severe combined immunodeficiency
SEM Standard error of the mean TGF-β Transforming growth factor β
TURP Transurethral resection of the prostate VEGF-C Vascular endothelial growth factor C
VEGFR-3 Vascular endothelial growth factor receptor 3 ZEB1 Zinc finger E-box binding homeobox 1
INTRODUCTION
The normal prostate gland
Anatomy and physiology
The prostate gland is a small, rounded organ with a diameter of approximately 4 cm. It is positioned immediately below the urinary bladder, where it encircles the proximal portion of the urethra. The prostate consists of glands, smooth muscles and connective tissue and is enclosed by a fibrous capsule-like structure. The glandular ducts open up into urethra. The human prostate can be divided into three distinct zones; the peripheral zone, the transitional zone and the central zone.1 The peripheral zone is the largest and the most common origin of prostate cancer.2 Studies of prostate cancer are often performed in rodent models. In contrast to humans, the rodent prostate gland consists of lobes; the anterior lobe, the dorsal and lateral lobe (collectively referred to as the dorsolateral lobe) and the ventral lobe.3,4 There is no clear analogy between the lobular structure of the rodent prostate and the zonal architecture of the human prostate. However, studies indicate that the dorsolateral lobe is most similar to the human peripheral zone.4
The prostatic glands produce a weakly alkaline secretion that contributes to about 30% of the semen. The secretion contains protein and ions and function as a liquefying agent that assists in sperm motility and its alkalinity protects the sperm in their passage through the acidic environment of the female vagina. The secretion is ejected into the urethra by peristaltic contractions of the muscular wall. The serine protease prostate specific antigen (PSA), also known as kallikrein III is perhaps one of the most well known secreted protein from the prostate.5 The prostate gland remains relatively small throughout childhood and begins to grow at puberty under the stimulus of testosterone. It reaches an almost stationary size by the age of about 20 years and remains mostly at this size up to the age of about 50 years. At that time, the prostate may start growing again, which sometimes leads to a state called benign prostatic hyperplasia (BPH).
Morphology
The glandular ducts are lined by a prostatic epithelium where three distinct cell types can be distinguished; luminal cells, basal cells and neuroendocrine cells (fig.1). The predominant cell type is the secretory luminal cell. Luminal cells are
terminally differentiated and characterised by the expression of the androgen receptor (AR).6 They produce PSA and prostatic acid phosphatase (PAP) and they are dependent on androgens for survival.7 The basal cells are relatively undifferentiated and they express low levels of AR8 but are not dependent on androgens for survival.6,7 They lack secretory function and expression of PSA.
Their function is not fully understood but it is believed that a subset of the basal cells function as stem cells in the prostate.9 It has been suggested that the androgen- independent (AI) prostate stem cells give rise to a population of androgen responsive transit amplifying cells that in turn can amplify the number of luminal cells.9,10 The characteristics of the transit amplifying cells are proposed to be intermediate between basal cells and luminal cells. Finally, the third prostatic epithelial cell type is the neuroendocrine cell, which are terminally differentiated and androgen-insensitive cells dispersed throughout the basal cell layer. They contain serotonin and thyroid-stimulating hormone that support the growth of the luminal cells.11 The stroma is composed of smooth muscle cells, endothelial cells, nerves, fibroblasts, dendritic cells and infiltrating immune cells. The fibroblastic stromal cells express AR and are androgen responsive.12-14 They produce growth factors for the epithelial cells in an androgen-dependent (AD) manner15 and the crosstalk between the stroma and epithelium is an important regulator of prostate growth and differentiation.16
Figure 1: Schematic illustration of the different cell types within the epithelium of a human prostate gland; luminal cells, basal cells and neuroendocrine cells.
Regulation of the prostate gland
Development and growth of the prostate gland is highly dependent on androgens.
The production of androgens is regulated from hypothalamus by secretion of gonadotropin-releasing hormone (GnRH), which acts on the pituitary gland. The pituitary responds with secretion of luteinizing hormone (LH), which thereafter induces the secretion of testosterone from the Leydig cells of the testis. In addition, the hypothalamus release corticotropin-releasing hormone (CRH) that induces the secretion of adrenocorticotropic hormone (ACTH) from the pituitary gland. ACTH influences the adrenal glands to produce testosterone and other weak androgens, for example adrenostenediol. Of the circulating testosterone, 95% originates from the testis and the remaining 5% originates from the adrenal glands (fig. 2).
Figure 2: The production of testosterone is under the superior control of the hypothalamus and the pituitary gland. The hypothalamus secretes GnRH and CRH that influences the pituitary to produce LH and ACTH, respectively. LH influences the testis to produce testosterone and ACTH regulates the production of testosterone and other weak androgens from the adrenal glands. The majority of the testosterone originates from the testis. GnRH = gonadotropin-releasing hormone;
CRH = corticotropin-releasing hormone; LH = luteinizing hormone; ACTH = adrenocorticotropic hormone.
Circulating testosterone diffuses into the epithelial and stromal cells of the prostate where it is converted by the enzyme 5α-reductase into dihydrotestosterone (DHT).
Both testosterone and DHT can bind the AR, but DHT has a stronger binding affinity and is more potent.17 Ligand-free AR in the cytosol is bound to heat-shock proteins (Hsp-70 and Hsp-90) that stabilize the receptor and protects it from degradation. Androgen binding to the receptor induces a conformational change that results in dissociation of the Hsp proteins. Two AR with bound ligand then form a homodimer that is stabilized by phosphorylation and transported into the
nucleus. Inside the nucleus the complex binds to target genes termed androgen response elements (AREs) and initiates transcription of genes regulating growth, differentiation and survival.
Effects of androgen deprivation
The normal prostate gland needs androgens for survival. Androgen withdrawal results in loss of secretory function, decreased cell proliferation and a rapid reduction in glandular size,18 which is caused by a widespread apoptosis among the epithelial cells.14,19 It was for a long time assumed that castration-induced epithelial cell death was mediated by decreased AR signalling in the epithelial cells.
However, recent studies indicate that it is in fact the stroma that regulates the major effects observed in the epithelium. Mice that expressed stromal AR but not epithelial AR responded similarly to androgen withdrawal as mice expressing AR in both stromal and epithelial cells. Mice that lacked AR in the stromal cells did not respond to castration at all.19 In addition, the prostate epithelial cell death is preceded by a major reduction in blood flow20,21 and by apoptosis of the endothelial cells.22 Similarly, testosterone administration results in the complete regeneration of the prostate gland, which is preceded by an increase in blood flow and regrowth of the vasculature.20,23 Castration-induced prostate involution is therefore partly caused by insufficient blood flow.
Prostate cancer
General background
Prostate cancer is one of the major health issues in the Western countries and for no other cancer form the incidence increases so quickly. Most likely, the increasing number of diagnosed cases originates from a frequent use of PSA as a diagnostic tool. Prostate cancer is the most frequently diagnosed cancer form in men in Sweden and approximately 9000 new cases are discovered each year. Prostate cancer also accounts for the most common cancer related death among men in Sweden, and each year about 2300 men die of the disease (Swedish Cancer Registry 2007).
The cause of prostate cancer is not known. Its occurrence is strongly related to age, and the majority of the patients are between 60 and 70 years when diagnosed.24 Prostate cancer is rarely diagnosed before the age of 50 years. There are large
geographic variations in the incidence, and prostate cancer is more common in Europe and in USA than in Asia, which points to the importance of lifestyle and environmental factors. It has been suggested that isoflavonoids, which are constituents of soy, have a protective effect against prostate cancer, which could explain the low incidence in Asia. It is also generally considered that lycopenes have a protective effect against development of prostate cancer. In contrast, high energy intake, high body mass index (BMI) as well as the metabolic syndrome are considered risk factors for prostate cancer.25 There is also an ongoing discussion if prostatic inflammation may contribute to the development and progression of prostate cancer. Some epidemiological studies have shown a significant association between prostatitis and prostate cancer26 while others have failed to demonstrate an association.27 Hereditary factors account for a relatively small fraction (5-7%) of the cases. In 1996, the first susceptible prostate cancer gene was discovered and it was named Hereditary prostate cancer 1 (HPC1). However, studies have shown that there are only a small fraction of the hereditary cases that are caused by HPC1 and hereditary at prostate cancer is probably complex.25
Prostate carcinogenesis
The cellular origin of prostate cancer is still controversial. It has been proposed that luminal cells are responsible for the tumor-initiating capacity, due to the fact that most prostate cancers display luminal characteristics. However, there is now increasing evidence that prostate cancer arises from the undifferentiated stem cells that are present in the prostate.9 Since the stem cells do not express AR and are independent of androgens, their presence is specifically interesting for the development of AI disease (see below).
Prostate cancer is a multifocal and heterogeneous disease, where several tumor locations are found within the prostate at the time of diagnosis.28 In general, prostate tumors are considered to be slowly growing.29 It is estimated that about half of the elderly male population have an insignificant, latent prostate cancer but the vast majority of them will never suffer from the disease. One problem today is to identify progression markers to distinguish indolent tumors from those tumors that will progress rapidly and cause the death of the patient.
The progression of prostate epithelial cells into a malignant phenotype is a multistep process (fig. 3). In many men, a precursor stadium to prostate cancer can be found, so called prostatic intraepithelial neoplasia (PIN). These premalignant lesions can develop into malignant tumors, which in the beginning are localized within the prostate. Additional genetic alterations result in development of a locally invasive tumor, which could break through the prostatic border and invade
surrounding tissue. Prostate cancer preferentially forms metastases to the bones and to the lymph nodes. As the normal prostate, prostate cancer is initially AD for growth and survival. This feature makes it possible to treat prostate cancer with androgen deprivation therapy (ADT). AD tumors respond to castration in a similar way as the normal prostate tissue, i.e. a reduction in blood flow and apoptosis of endothelial cells30 and of epithelial tumor cells,31,32 which altogether results in reduced tumor burden. However, the initial androgen-dependency is generally lost during tumor progression and after a certain time an AI tumor relapses. AI tumors (or hormone-refractory tumors) are highly aggressive and metastatic and the patient survival is generally less than a year.
Figure 3: The progression of prostate cancer is a multistep process.
(PIN = prostatic intraepithelial neoplasia).
Diagnosis and pathology
The common method to diagnose prostate cancer is through rectal palpation or transrectal ultrasound together with core biopsies from the prostate. In addition, the PSA test is commonly used to assess the risk for prostate cancer. Normal prostate tissue prevents PSA to reach the blood. However, in the diseased prostate (i.e.
prostate cancer and prostatitis), the basement membrane is leaky, resulting in increased blood levels of PSA. A normal PSA value is in the range of 0-3 ng/ml.
However, many men over 50 years have a PSA value between 3 and 10 ng/ml, which could be due to prostate cancer, but more often this is due to BPH.
Unfortunately, the PSA test has low specificity and sensitivity and therefore it is not possible due to PSA to differentiate between low and high malignancies. If the PSA value is above 100 ng/ml it is generally an indication of widespread metastatic disease.25 After confirmed prostate cancer diagnosis, additional investigations with regard to metastases are performed, which include radionuclide bone scans to detect possible metastatic lesions in the bone. In addition, in relation with prostatectomy, regional lymphadenoectomy could be performed to investigate for presence of lymph node metastases with subsequent histology. Prostate tumors are usually classified according to the TNM system, which makes it possible to study prostate tumor stage over time as well as prognosis for the individual patient (table I).
Table I: TNM classification of prostate tumors T – Primary tumor
TX Primary tumor cannot be assessed T0 No evidence of primary tumor
T1 Clinically unapparent, neither palpable nor visible by imaging T1a Tumor incidental histologic finding in 5% or less of tissue resected T1b Tumor incidental histologic finding in more than 5% of tissue resected T1c Tumor identified by needle biopsy
T2 Tumor confined within the prostate T2a Tumor involves half of a lobe or less
T2b Tumor involves more than half of a lobe but not both lobes T2c Tumor involves both lobes
T3 Tumor extends through the prostatic capsule T3a Unilateral extracapsular extension T3b Bilateral extracapsular extension
T3c Tumor invades seminal vesicles
T4 Tumor is fixed or invades adjacent structures other than the seminal vesicles:
bladder neck, external sphincter, rectum, levator muscles, or pelvic wall N – Regional lymph nodes
NX Regional lymph node cannot be assessed N0 No regional lymph node metastases N1 Metastases in regional lymph nodes M – Distant metastases
MX Distant metastases cannot be assessed
M0 No distant metastases
M1 Distant metastases
The most common way to obtain tissue specimens from the prostate is with needle biopsies through the rectal wall. There is no standard direction on how many biopsies that should be sampled and how they should be taken. However, usually between 6 and 12 biopsies are taken. The Gleason score33 is the most commonly used histological grading system for prostate cancer and it correlates with tumor progression.34,35 This method was established in the 1960s and is based on the growth pattern of the tumor. The Gleason score is the sum of the most common and the most aggressive growth patterns that are graded from 1 to 5, with 1 being the least aggressive and 5 the most aggressive.
Treatment of prostate cancer
Treatment of localized prostate cancer
Localized prostate cancer is the most commonly diagnosed cancer stage. The choice of treatments is active surveillance, radical prostatectomy or radiotherapy and is based on the patient’s life expectancy and grade of malignancy. The recommendation is that men with a short life expectancy who have early stage prostate cancer should be followed by surveillance as first choice.36 Younger patients with longer life expectancy or patients with more poorly differentiated tumors are offered curative treatment. The most common curative treatment in Sweden is removal of the prostate gland with radical prostatectomy. Another curative treatment option for localized prostate cancer is radiotherapy.36
Treatment of locally advanced prostate cancer
In patients with extracapsular tumor extension, prostatectomy is not the first option, since there can be difficulties to remove the whole tumor. Treatment recommendations for locally advanced prostate cancer are instead a combination of hormonal therapy and dose-escalating radiotherapy.37 Another option is hormonal treatment in the form of anti-androgens or castration therapy (see below).
Treatment of metastatic prostate cancer
Metastatic prostate cancer is treated by ADT. Already in 1941 Huggins and coworkers performed their pioneering work on hormonal treatment of advanced prostate cancer,38 which was later awarded with the Nobel Prize. Testicular androgens can be eliminated by surgical or medical castration. Medical castration includes treatment with GnRH analogs that exerts its pharmacological action through downregulation of the GnRH receptors present on pituitary gland. This results in inhibition of LH and the subsequent testosterone secretions from the Leydig cells of the testis. GnRH agonists decrease serum testosterone to castration levels after 3-4 weeks. This is preceded by a transient increase in serum testosterone, known as the flare period, which could cause worse symptoms for the patient.36,39 To avoid the flare period, GnRH antagonists that have direct actions on the receptors, have recently been developed.40 Chemical castration can also be obtained by administration of estrogens, which results in decreased secretion of LH from the hypothalamus via a negative feedback loop. In addition, anti-androgens, such as bicalutamide, can be used to block the androgenic effects. Anti-androgens have peripheral actions by directly binding and blocking the AR in the prostate cancer cells.36,39
About 80% of the patients will have symptomatic relief after ADT.36 Although endocrine therapy is palliative and not biologically curative, this treatment could
contribute to the decline in mortality rates by delaying death from prostate cancer long enough for the patient to die of unrelated causes. From animal experiments and clinical trials it has been shown that early initiation of endocrine therapy is beneficial, at least in more aggressive cancers, and improves survival.41,42 Intermittent androgen ablation is a treatment modality that has been introduced into the clinic43 but there are no relevant endpoints yet, therefore this should so far be considered as experimental.
Treatment of AI prostate cancer
Treatment of AI prostate cancer is only palliative and median patient survival is less than a year. Despite the AI nature of the tumor, it is of importance to continue with ADT in this stage of disease.44 In addition, a second-line treatment with anti- androgens can be beneficial for the patient. Treatment with the cytotoxic drug docetaxel has been introduced in the clinic and it has been shown to improve median survival with a few months.45,46 Prednisone is also used in the treatment of AI prostate cancer and it often results in improved well-being. Development of new treatment strategies for the AI stage of prostate cancer is of importance and is urgently needed. There are several drugs that are in early clinical trials.47 However, there are no conclusive results yet and the role of these drugs for treatment of prostate cancer will be proven in the future.
Mechanisms of development of AI prostate cancer
ADT results in a temporarily relief for patients with advanced prostate cancer, but will eventually trigger the development of an AI prostate tumor. AI prostate cancer is highly aggressive and metastatic and is a major challenge for clinicians. The factors that trigger development of AI prostate cancer are currently not known.
Neither is the point when the molecular alterations that promote AI prostate cancer occur. Results from an early study suggested that untreated metastatic tumors already contain the alterations needed for recurrence to occur during the pressure of ADT.48 However, later studies instead support the theory that ADT trigger the molecular alterations that result in development of an AI prostate cancer.49,50 Recurrent prostate tumors often reexpress AR target genes and AR is observed at high frequency in these tumors.51-53 In addition, AI tumors respond to additional hormonal manipulations, such as anti-androgens.54 These data suggest that most of the recurrent tumors are neither AI nor hormone-refractory, because they continue to depend on the AR signaling axis for growth. A more accurate name for AI/hormone refractory prostate cancer is therefore “castration-resistant” prostate cancer. Transition into androgen-independency is a complex process and despite a lot of effort in resolving this issue, the detailed molecular mechanisms remain
unknown. Theories to the development of AI prostate cancer includes several AR related mechanisms but also mechanisms not related to the AR.
AR related mechanisms
Much attention has been paid to the AR in the development of AI prostate cancer.
Several mechanisms that serve to activate the AR in absence of androgens have been described (fig. 4).
1) Amplification of the AR gene could be one possible mechanism that facilitates proliferation of prostate cancer cells in low concentrations of androgens. Visakorpi and colleagues reported that AR amplifications are increased during ADT,55 which has later been confirmed by others.52,56 Amplifications have also been observed in bone metastases57 suggesting the involvement of AR in the progression of the disease.
2) Point mutations in the steroid binding domain of the AR are another mechanism that has been proposed to be of importance in development of AI prostate cancer. Mutations result in a promiscuous receptor that allows non- specific binding of ligands such as estrogens and non-steroidal anti- androgens. The frequency of AR mutations increases in the AI stage58,59 and has also been observed in lymph node and bone metastases.60 Veldscholte et al was the first to describe that the LNCaP cell line harbor a AR mutation in the ligand-binding domain, thus allowing it to be activated by other ligands than androgens.61
3) Hypersensitivity of the AR is another mechanism that can drive the proliferation of the prostate cancer cells under low androgen concentrations.
Gregory et al have reported that this mechanism includes increased stabilization and increased nuclear transportation of the AR.62,63
4) Coactivators interact with steroid receptors and enhance their ligand- dependent transactivation. Some examples of co-activators that have been reported to be upregulated in AI prostate cancer are androgen receptor associated protein-70 (Ara70), steroid-receptor coactivator-1 (SRC-1), transcriptional intermediary factor-2 (TIF-2) and receptor-associated co- activator-3 (RAC3).62,64,65
5) Androgen deprivation can also result in activation of intracellular signaling transduction pathways that can drive the proliferation of the prostate cancer cells instead of androgens. These pathways can facilitate activation of the AR in absence of androgens or the other alternative is that the AR is bypassed altogether. It has for instance been demonstrated that AR can be
activated by insulin-like growth factor 1 (IGF-1), the keratinocyte growth factor (KGF) and the epidermal growth factor (EGF).66
Figure 4: Mechanisms that have been suggested to be responsible for development of AI prostate cancer. Theories include several AR related mechanisms (1-5) but also mechanisms not related to the AR (6). AR = androgen receptor; DHT = dihydrotestosterone; PSA = prostate specific antigen.
Other mechanisms
6) Development of androgen-independency can also be explained by the upregulation of the anti-apoptotic protein Bcl-2 (fig. 4). Bcl-2 is not normally expressed by the secretory prostate epithelial cells.67 However, PIN lesions frequently express Bcl-2 and increased levels are observed in AI specimens.67 Correspondingly, Bcl-2 is upregulated after castration in experimental models of prostate cancer68 and inhibition of Bcl-2 resulted in delayed progression to androgen-independency.69 Androgen deprivation therefore seems to induce signals that results in bypassing of the apoptotic response.
7) Prostate tumors are extremely heterogeneous and are probably comprised of tumor cells with varying degree of androgen-sensitivity and responsiveness.
One theory to the development of AI disease is the clonal expansion theory.9 Coffey and Isaacs have suggested that androgen-independence is due to the existence of AI prostate cancer stem cells in the original population of prostate cancer cells.70 Androgen deprivation would result in depletion of the
AD cells but promote disease progression by activating normally quiescent cancer stem cells that would repopulate the tumor with AI cells. Craft et al provided further support of the clonal expansion theory and showed that AI cells account for 1 in 105 cells even before initiation of androgen ablation.71
Interactions between cancer cells and stroma
Earlier studies by Cunha and coworkers have revealed that the normal prostatic stromal cells control the differentiation and development of the normal epithelial cells and prostate gland.72 In a similar way, there exists a continuous communication between prostate tumor cells and stromal cells. Several biological experiments have provided profound evidence that stromal cells play an important role in prostate cancer initiation and progression. When non-tumorigenic prostate epithelial cell lines were combined with carcinoma-associated fibroblasts (CAFs) and implanted as xenografts in immunodeficient mice, tumors were established.
Tumors were not formed when epithelial cells were injected alone,73-75 showing the importance of CAFs in tumor initiation and progression.
The reciprocal interactions between stroma and cancer cells are not only promoting growth of the cancer cells. The cancer cells also have a profound influence on the stromal cells and the stroma in the vicinity of the tumor is often referred to as
“reactive stroma”. Experiments addressing the issue of differences between CAFs and normal stroma have shown a dramatic difference in tumor forming capacity of non-tumorigenic epithelial cells mixed with cancer stroma or normal stroma.75 The CAFs promoted a rapid tumor development and progression, while normal stromal cells restricted tumor formation. Therefore, during the course of tumor development there are profound changes in the stroma, helping the neighbouring cancer cells to survive, proliferate and ultimately forming metastases.76
Although androgens are important for maintenance of the normal prostate, they are not alone responsible for the regulation. There are several growth factors identified that promote interactions and communications between the stromal and epithelial compartments. The major prostatic growth factor families include fibroblast growth factor (FGF) family, transforming growth factor-β (TGF-β) family, IGF family and EGF family. During the course of prostate tumorigenesis many of the normal prostate growth factors are altered.77 Exactly how the reactive stromal cells influence and regulate the tumorigenic process are not well defined. It is most possible that the interactions between the stromal cells and cancer cells differ depending on tumor stage. Furthermore, because there are phenotypic differences between AD and AI prostate cancer cells it would be interesting to reveal differences in their respective communications with the stromal cells.
Invasion and metastasis
The metastatic process
Primary tumors impair normal tissue function. However, they are only responsible for about 10% of the deaths from cancer. The remaining 90% of the cancer deaths are caused by metastatic disease.78 Metastases create a great chaos throughout the body and therefore they are the most dangerous manifestations of the cancer process. For unknown reasons, certain types of tumors never metastasize, while others have a high probability to do so. For instance, prostate cancer has a high propensity to form metastases to the regional lymph nodes and to the bones.
There are a number of sequential steps that a cancer cell must overcome to succeed in the establishment at a new site (fig. 5).79,80 The first step is the detachment of single cancer cells from the primary tumor, which involves alterations in the cell adhesion profile. To gain access to the vessels for further transportation, the tumor cells must break through the basement membrane and path their way through the tissue. The invasive properties of tumor cells enable them to move through the vessel wall and enter into the circulation, a process called intravasation. Once in the circulation, individual cancer cells may travel with the blood or lymph flow to distant sites. However, the blood represents a hostile environment for metastasizing tumor cells. Hydrodynamic shear forces are present in the blood, which may tear the cancer cells apart.78 Experimental studies have shown that survival is greatly enhanced if the tumor cells can attract platelets to escort them through the rapid blood flow.81 In addition, the most common way for invading cancer cells to move through the tissue is as a unit with other cancer cells, thus enhancing their survival in the circulation.78 Further, once the tumor cells enter the circulation they will lose their anchorage to the underlying stroma. Like normal cells, cancer cells may continue to depend on solid substrates for survival. Many cells will therefore rapidly die as a result of anoikis (apoptosis that is triggered by detachment from a solid substrate).82
The next step for the metastasizing tumor cells that have managed to survive in the circulation is to invade the new organ by adhering to the vessel wall and penetrate into the surrounding tissue, a process called extravasation. At this point, tumor cells use two alternative options. Either they can start proliferate inside the vessel, creating a small tumor that pushes on the vessel wall and forces the endothelial cells to separate so the tumor can pass. The other option is to invade the endothelial wall directly and start proliferating at the new tissue.78 The last step, called colonization, is perhaps the most challenging. Many metastasizing tumor cells will die once they have arrived at the new site, while others will remain dormant for many years.83 In general, the number of micrometastases in a cancer patient vastly
increases those that will eventually expand in size.78 To be able to expand, the cancer cell must initiate the formation of new blood vessels, a process called angiogenesis. Oxygen and nutrients can only diffuse 1 mm in distance and therefore newly formed metastatic foci cannot grow over a size of 1-2 mm3 without initiating angiogenesis.84
Figure 5: The metastatic cascade consists of several interrelated, sequential steps. 1) Tumorigenesis. 2) Angiogenesis. 3) Detachment of tumor cells, invasion through the tissue and intravasation into vessels. 4) Transport through the circulation. Tumor cells must survive the hydrodynamic shear forces that are present in the circulation. Survival is greatly enhanced if the tumor cells form aggregates with each other or with platelets. 5) Arrival at the new site. For prostate cancer this is often the bone. 6) Tumor cells adhere to the vessel wall. 7) Extravasation into the new organ parenchyma. 8) Colonization, initiation of angiogenesis and subsequent proliferation. Experimental studies have shown that the last step is the rate-limiting.
Overall, metastasis is an inefficient process. Of the several millions of tumor cells that are seeded into the circulation it is only a small fraction that successfully complete all the steps in the metastatic cascade.85,86 Experimental studies have lead to the conclusion that the initial steps in the metastatic cascade are completed easily for most tumor cells.87,88 However, the last step involving the colonization and initiation of angiogenesis at the new site is the rate-limiting. It has been reported that three days after intraportally injection of melanoma tumor cells, 83% of the cells had extravasated but only 2% of the cells formed micrometastases.
Furthermore, 0.02% of the cells persisted and developed into lethal metastases.88 Similar results have been obtained by others.89
In order to acquire motility and invasiveness, the prostate cancer cells must shed many of their epithelial traits and undergo drastic alterations; the epithelial- mesenchymal transition (EMT). EMT involves changes in cell morphology and gene expression pattern, resulting in gain of mesenchymal characteristics. These alterations results in extended and elongated cancer cells, allowing for increased migratory capacity.90 In addition, there are major changes in the gene expression profile during EMT. Expression of E-cadherin and cytokeratins are repressed and instead mesenchymal markers, such as N-cadherin, vimentin and fibronectin, are induced.78
The seed and soil theory
It has long been recognized that different cancer types show an organ-specific pattern of metastasis. For instance, prostate cancer preferentially forms metastases in the bones and metastases from breast cancer are often detected in the brain, lungs, bone and liver. Already in 1889, Stephen Paget published a paper that described the seed and soil theory.91 He had noticed that certain types of tumor cells (the seed) had a high propensity to form metastases in specific organs (the soil) and he proposed that this was due to the compatibility between the seed and the soil.
This idea was challenged in 1920s by James Ewing, who suggested that the patterns of blood flow were the primary reason for organ-specific metastasis.79 Today we know that both models are valid. Autopsy studies and experimental animal studies support the concept that both blood flow and compatibility factors contribute to metastatic spread to various organs.79 Together, these studies show that the blood flow determine the initial fate of the tumor cells and decide in which organ the tumor cells will end up after they have left the primary tumor. However, after the cancer cells have arrested in an organ, their ability to grow there is dependent on molecular interactions between the cancer cells and the new environment. In addition, the new organ must be able to support the cancer cells with the proper growth factors. Because chemokines and their receptors are involved in the homing of lymphocytes and hematopoietic cells to specific organs, it could be reasoned that also cancer cells use chemokines to home to specific organs. In an elegant study by Muller and coworkers it was shown that breast cancer cells express the chemokine receptors that match the chemokines that are expressed in the organs where these cells end up as metastases.92
Tumor dormancy
It has been observed that metastatic relapse of a tumor can occur decades after removal of the primary tumor.93 It has therefore been hypothesized that primary tumors shed metastatic cells to the circulation in an early phase of the disease.
These scattered tumor cells can persist in an inactive state for many years, called tumor dormancy.83 Which mechanisms that awake the cancer cells are at present unknown. Studies that have modeled tumor dormancy mathematically indicate that continuous slow growth is unlikely. Instead a model favoring discontinuous growth with periods of quiescence is more likely.94,95 Judah Folkman and colleagues could show the existence of preangiogenic metastatic foci in the metastatic niche. In these preangiogenic metastases the proliferation was counter balanced by apoptosis, resulting in no net increase in tumor volume. When these foci gained the ability to vascularize, tumor dormancy ceased and cancer cells started to proliferate.96 Another possible contributor to tumor dormancy is presence of solitary cancer cells at the metastatic site.97 Many cells that arrive in the secondary site fail to initiate cell division but remain as quiescent cancer cells. It has been shown that recovered solitary mammary carcinoma cells from liver tissue, retained their tumorigenic capacity when re-injected into the mammary fat pad of mice.97 These experiments show that despite their apparent dormancy at the secondary site, the cells are still active. A better understanding of tumor dormancy and the molecular factors that contribute to the subsequent initiation of cell division is important to be able to treat metastatic disease.
Lymphangiogenesis and metastasis
The lymphatic system
The involvement of the lymphatic system in the metastatic process has been intensively investigated over the last years. Many carcinomas, including prostate cancer, metastasize to sentinel lymph nodes via the lymphatic vessels. Therefore, sign of lymph node metastases is often used as cancer staging in the clinical setting.
Presence of cancer cells in the lymph nodes are considered as an adverse prognostic advent in many carcinomas.98 Thus, removal of the regional lymph nodes is often standard procedures, resulting in improved patient survival.
Lymphatic vessels resemble blood vessels, but are generally thinner. The lymphatic endothelium has poorly developed junctions and large interendothelial gaps, which results in a relatively free import of interstitial fluid. The lymphatic capillaries also lack a continuous basement membrane and are devoid of pericytes. These properties make lymphatic vessels susceptible for invasion by tumor cells.
Traditionally, the lymphatic system has been considered to be passively involved in
the metastatic process. However, studies have demonstrated formation of new lymphatic vessels in tumors, so called lymphangiogenesis,99-101 which is evidence of lymphatic vessel activation in cancer. The detailed molecular mechanisms behind lymphangiogenesis have been poorly understood but have improved after the identification of specific lymphatic growth factors and lymphatic endothelial markers. The specific lymphatic endothelial markers include lyve-1,102 podoplanin103 and prox-1.104
Vascular endothelial growth factor C (VEGF-C)
The first lymphangiogenic growth factor was isolated in 1996 from human prostatic carcinoma cells.105 It was found to belong to the group of vascular endothelial growth factors (VEGFs) and it was named VEGF-C. Later studies identified another family member, VEGF-D106 that also stimulated growth of lymphatic vessels. VEGF-C and VEGF-D are glycosylated, secretory proteins and by means of proteolytical processing several forms are generated. They mediate their effects on lymphatic vessels through the VEGF receptor 3 (VEGFR-3). The expression of VEGFR-3 was originally thought to be restricted to lymphatic endothelial cells107 but further studies revealed that VEGFR-3 was also expressed by a small subset of blood vessels.108,109 By proteolytic processing, VEGF-C and VEGF-D gain the ability to bind to the VEGFR-2 that is expressed on blood vessels. Thus, they can have actions on blood vascular endothelial cells and induce angiogenesis.110 VEGF- C is definitively one of the main lymphangiogenic growth factor but since its discovery others have been identified, including PDGF-BB,111 FGF-2112 and the angiopoietin-1 (Ang-1) and -2 (Ang-2).113
The expression of VEGF-C mRNA is upregulated by different factors, including PDGF, EGF, TGF-β and also by serum.114 In addition, it has been demonstrated that androgen deprivation induces expression of VEGF-C in prostate cancer cells.115 In contrast, VEGF-C is not regulated by hypoxia, RAS oncoprotein or mutant p53, which are potent inducers of VEGF expression.114 In normal human adult tissue, VEGF-C is expressed most prominently in the heart, placenta, ovary and in the small intestine.105 Macrophages are another important source of VEGF- C.116
VEGF-C and its role in lymphatic metastasis
Today, there is strong evidence that lymphangiogenesis does occur in the presence of VEGF-C (fig. 6). VEGF-C can induce hyperplasia of preexisting lymphatic vessels117 and can also stimulate the proliferation of newly formed lymphatic vessels.118 Overexpression of VEGF-C in transgenic and xenograft models of human cancer induced lymphangiogenesis around the tumors and enhanced metastatic spread of cancer cells to regional lymph nodes.119-122
VEGF-C is upregulated in many types of human malignancies and is also related to the appearance of lymph node metastases.123-126 Previous to our publication (paper I), Tsurusaki et al reported on elevated mRNA levels in tumors from patients with lymph node metastases in comparison to tumors from patients without lymph node metastases.127 Similarly, expression of VEGFR-3 by lymphatic endothelial cells has been shown to be associated with lymph node metastasis in prostate cancer.128
Figure 6: Lymphangiogenic growth factors (VEGF-C, VEGF-D, PDGF-BB, Ang-1, Ang-2, FGF-2) are secreted from the tumor cells and induce the formation of new lymphatic vessels.
VEGF-C/D is the most studied lymphangiogenic growth factors and they mediate their effects via VEGFR-3 that is expressed on the lymphatic endothelial cells. Increased lymphatic vessel density in tumors is associated with presence of regional lymph node metastases. Both peritumoral and intratumoral lymphatic vessels have been observed in tumors but there are doubts if the intratumoral lymphatic vessels are functional.
More recently, is has been observed that lymphangiogenesis not only occur in the vicinity of the primary tumor but also in the sentinel lymph nodes.129,130 Notably, lymph node lymphangiogenesis was observed even before the arrival of tumor cells, which indicate that the primary tumor starts to prepare the metastatic site prior to the dissemination of tumor cells.131,132 Although metastasis to distant organs most likely occurs via the hematogeneous system, it might require the initial spread of tumor cells to the regional lymph nodes, from where the cancer cells can spread further. This is supported by a study where metastasis to distant organs was not observed without simultaneous lymph node metastasis.129 Correspondingly,
human lymph nodes infiltrated with metastatic melanoma cells also exhibited lymphatic vessel growth, indicating that lymph node lymphangiogenesis can be a feature of human cancer.133
Cell adhesion molecules in cancer
Cell adhesion molecules (CAMs)
Cell adhesion molecules (CAMs) are present between cells and between cells and the extracellular matrix (ECM) and they constitute key components that maintain the normal structure, integrity and function of cells and tissues.134 Epithelial cells connect to each other and to the ECM by adherens junction, desmosomes and cell- matrix adhesion complexes (fig. 7). The adherens junctions, consisting of the cadherin class of CAM, connect to the cytoskeletal actin filaments, which creates an adhesion belt around the cell. The desmosomes on the other hand, connect adjacent cells via the intermediate filaments. Cell-cell adhesion is mainly mediated via the cadherins and cell-matrix adhesion is mainly mediated via the integrins. In addition, there are also adhesive proteins that are part of the tight junctions. Tight junction proteins maintain the polarity of epithelial cells by separating the apical side from the basolateral side of the cells. During cancer development, there are major rearrangements and alterations in the CAMs, resulting in increased tumor cell motility.
Figure 7: Epithelial cells are connected to each other via adherens junctions and desmosomes.
The adherens junctions provide cell-cell adhesion via the cadherins, which is connected to the actin cytoskeleton that form an adhesion belt around the cell. The desmosomes connect to the intermediate filaments. Tight junctions separate the apical side from the basolateral side, keeping the epithelial cells in a polarized state. Cell-matrix adhesion is mainly mediated via integrins.
ECM = extracellular matrix.
Cadherins
Structure and function
The cadherins are a superfamily of CAMs that mediate adhesions between cells in the presence of extracellular calcium. Different cadherins are differentially expressed during embryonic development, indicating distinct functions in cell adhesion. The cadherins are composed of a large extracellular domain that binds with homophilic bonds to cadherins on neighboring cells, resulting in stable forces between adjacent cells. Cadherins also have one transmembrane segment and a highly conserved intracellular domain. The cytoplasmic part of the cadherins is anchored to the actin cytoskeleton via a second group of proteins called the catenins.134 The linkage to the cytoskeleton is crucial for the cell adhesion function of the cadherins. There are three major catenins; α-, β-, and γ-catenin that regulate the function of the cadherins.135 The most extensively studied cadherins are the three classical cadherins E-cadherin, N-cadherin and P-cadherin.
E-cadherin and cancer
E-cadherin maintains the integrity and polarity of normal epithelial tissue. Over the last years, several investigations on the functional role of E-cadherin have been performed. Inhibition of the function of E-cadherin with blocking antibodies resulted in disruption of cell contacts in vitro and induction of a more motile phenotype.136-138 Conversely, forced expression of E-cadherin in cancer cells impaired the invasive capacity.137,139 These results clearly demonstrated a critical role for E-cadherin in tumor invasion. Furthermore, using the Rip1Tag2 transgenic model of pancreatic cancer, Perl et al could establish a casual role between loss of E-cadherin and transition from adenoma to carcinoma.140 Therefore, loss of E- cadherin results in disruption of the tight contacts that exist between adjacent epithelial cells allowing single metastatic cells to escape from the primary tumor. In most cases, loss of E-cadherin is due to transcriptional repression by hypermethylation or chromatin rearrangements in the promoter region.141-144 Recent reports have also highlighted the role of snail, slug, twist and ZEB1 in silencing of the E-cadherin gene.145-148 These transcription factors act as repressors of E- cadherin gene expression, and thus inducing the metastatic phenotype. In addition, loss of E-cadherin can also be a consequence of mutations of the E-cadherin gene149 or aberrant tyrosine phosphorylation of E-cadherin and associated proteins by receptor tyrosine kinases,150,151 which result in disruption of E-cadherin mediated adhesion.
Reduction in E-cadherin not only results in decreased cell adhesiveness that is important for the first step in the metastatic cascade. It also affects signaling transduction pathways that can influence later steps in the metastatic cascade.152,153 A critical intracellular event resulting from loss of E-cadherin is the accumulation of free β-catenin in the cytosol. Besides for being an important component of the
